Re: object algebra

From: Neo <neo55592_at_hotmail.com>
Date: 4 Mar 2004 12:04:26 -0800
Message-ID: <4b45d3ad.0403041204.282f77a9_at_posting.google.com>


> > You're confusing missing information and nulls.
> > RDM creates NULLs when information required by a tuple is missing.
>
> Can you illustrate? Which operator(s) create(s) nulls? Are you sure
> that operator is in the TTM version of the RDM? Which version of the
> RDM are you assuming - or, if you prefer, whose exposition of the RDM
> are you using?

The creation of NULLs in RDM is rooted in its definition of a relation. The definition which optimizes RDM for its scope is the very same definition that causes NULLs when RDM is applied beyond its scope. Following is a poor analogy: RDM is used in an application where are sending a Rover to Mars (unknown reality) to collect some things (data). NASA scientists (designers) put there heads together and determine a 3-cubicle container will be required (relation heading). When the Rover lands on Mars, it constructs a 3-cubicle box (tuple) as pre-planned at design-time but only finds a rock, a bone, but no flower. The third cubicle remains empty (NULL). The Rover fills the empty cubicle with peanuts (mask for NULL ie N/A).

Like Newtonian, RDM has a scope within which it matches reality. Like Newtonian, when RDM is pushed beyond its scope, it incurrs NULLs and can provide answers that don't match reality. To push RDM to the edge of its limits, try implementing the simple example at www.xdb1.com/Example/Ex076.asp . Data should be NULL-less and normalized (but don't normalize down to symbols. It is impractical in RDM.) You don't have to model the realtors, isa, is and obeys. They can be assumed. Below are the base tables that one might start with.

T_CmdHier
T_God
T_ArmedForce
T_Church
T_Person
T_Dog
T_Computer
T_Age
T_Weight
T_Color
Received on Thu Mar 04 2004 - 21:04:26 CET

Original text of this message